apologies for double post. Dunno what happened, was trying to nuke the horrifically outdated sig via editing the post.

Something I forgot to mention: I'm assuming a single-roll combat table. If yellow attacks are on a double roll, this makes +hit a bit better, I'll model it but I don't think it's huge. If white attacks are on a double roll I'll be very surprised.

I'm afraid this doesn't make any sense, it either increases your DPS by a useful amount, in the context of your current stats, or it doesn't.

That doesn't follow at all. Suppose for argument's sake you do 1000 'potential' white damage (every thing connects) over a period of time, have a 35% crit rate, and don't glance (level 70 mob). 46% of your white damage is coming from hits, and 64% from crits. The attacks you miss by not having any +hit are hits, not crits, so you're losing 8.6%/65% = 13.2% of your *hit* damage, i.e. 13.2% * 46% = 6.1% of your total damage.

And this doesn't apply at all to yellow attacks unless you assume that you don't have any means/time to convert the 82% of energy you get back into DPS.

Time is not a scarce resource in most situations for cats. If wewere comparable to warlocks, you'd be running dotimer to refresh feral faerie fire at the very last second, like moonkins do.

I do in fact model this.

I'm not sure what all the math is about. +1% to hit is +1% of swings that are eligible for an OOC proc. As mentioned, it's a small increase to the power of +hit.

Yes it is, you're overestimating lost energy due to misses by a factor of 5. Lost combo points are a small effect because the misses cost you 18% of energy. And yes I model this

Dodge doesn't have anything to do with +hit other than modifying your base hit-rate- i.e. the point at which you take your derivative.

10%? 36%? what?

That's the conclusion I'm coming to (although FCS seems to be a lot better than that for the first 20 rating). My question then is, if hit is worse than pretty much any other DPS stat, then why does everyone seem concerned with what the hit cap is, when optimal gear shouldn't be anywhere near it?

Definitely, +hit makes your cycles much more predictable and therefore almost always better. A simplified, resilient cycle helps with the modeling though. The cycle I'm using for calculation at the moment has things like 'shred to 5CP', and the amount of energy~time taken for this (accounting for misses) is used to calculate the proportion of shreds mangled. In practice if you missed four shreds in a row you'd probably mangle again at some point before rip. Of course the resulting numbers only approximate those for a different cycle, and there's some unlikely corner cases that break the model.

Pretty sure anyone who makes a spreadsheet ends up waking up in the middle of the night thinking 'what if I wait for 52 energy so I can mangle rip but mangle misses' or 'what if the rogue's 1.3 speed green offhand rolls low and glances, winding up 100 below the mob's block - can glances even block?!', their psychiatrist tells them to make a monte carlo simulator,and they're never heard from again.

If it costs 29% less and does 50% less, then it's just worse.

This is very true. However I can't find any current theory of glancing blows that could put 1% to hit as 1% or more of your white DPS.

Not true, +hit takes a 0 multiplier to a 1.0 multiplier, +crit takes a 1.0 multiplier to a 2.2 multiplier (predatory instincts), 2.27 if you use a relentless earthstorm meta. Hit has advantages for things that proc on hit (OOC), crit is better for things that proc on crit (Hourglass).
But yes, point-for-point hit is better than crit if only looking at white damage.

What FCSR do you currently have? In my current gear set with Shapeshifter's signet equipped (20FCSR = 5FCS), FCSR is worth about 2.2AP against bosses. If I take off my signet, FCSR becomes worth 2.93, as the first 20 points are more valuable. (I'm looking at the positive derivative because I'm generally looking at upgrades not downgrades

Agi and str seem to be balanced somewhere in the mid-2k-ap gear range (depending on other stats), and change very slowly relative to each other, I'm happy to take them as interchangeable at my level of gear.

So I'm not all talk, here's my current (possibly broken) version

HTML (read only)
XLS
ODF

The current version gives 1 hit rating as 1.40AP for my gear, a substantial improvement (and i think, more accurate, thanks guys!). It's still the worst DPS stat though (I don't have haste modeled).

Roughly speaking, you type in the yellow fields.
Sheets depend on sheets on their left generally (exceptions: damage section of attack depends on cycle, pawn section of stat value depends on gems).
Calculating value for each step is hacky atm since TABLE() isn't supported - you enter a 1 in the STR column, it increases your STR by 10, you type the DPS into the box. To get AEP values you have to do this for base (no modifiers), AP, and the stats of interest.

Now I find what I lied about modelling.

These results are almost exactly the same as the results Emmerald's Spreadsheet gives except for FCS.

How exactly are you modelling FCS? I'd assume its the exact inverse of defense skill except for the part where it adds extra crit to mobs above your level. I was unclear on how exactly it does that (as are many from the posts I've read).

Reassuring .
Now here's a silly question - where can I actually find the thing?!
All I can find is links to the static gear listing, but I've got no idea what the base set of gear for this is. (If you have 2400AP and 40% crit then your stat valuation for upgrades won't be the same as if you have 2900AP and 30% crit).

It gives weapon skill, which reduces miss chance (8% + 0.04 * (mob_def - wpn_skill)).
When mob def exceeds your weapon skill, it gives a bonus 0.1% hit per skill.
When mob def exceeds your weapon skill + 10, it gives an additional 0.1% hit per skill.
So on a boss, the first 5 skill give 0.2% + 0.04% hit each. The next 10 give 0.1% + 0.04%, and any others give 0.04%.
I need to find my reference for this when I get home.

FCS gives weapon skill, which increases crit chance (agi_crit + critrating_crit - 0.04*(mob_def - wpn_skill) + talents)
When mob def exceeds your weapon skill, it gives a bonus 0.1% crit per skill.
So on a boss, the first 15 skill give 0.1% + 0.04% crit and the rest give 0.04%. (This was broken in the spreadsheet, you'd get the 0.1% for all skill. Fixed!)
This was in the 1.13 patch notes.

So all this could be completely wrong. This seemed at the time I looked into it to be the most likely/consistent model, but I don't think anyone knows for sure =\

http://www.gurgleblaster.net/forum/i...hp?topic=167.0 had the code for Emmerald's Sheet but the spreadsheet was removed recently. Not sure if its being added back.

http://forums.worldofwarcraft.com/th...sid=1&pageNo=1 has the link to Lolaan's spreadsheet, though it seems to be giving inflated values for Agi and Crit. Unsure if it has to do with gear setup or what though.


Ok the crit part seems consistent with what the patch notes said ("The player will gain 0.1% to their critical strike rating per weapon skill against monsters above their level.") though it doesnt specify that it stops at 15 points since it simply says mobs above your level (kinda academic since getting more than 15 weapon skill is not optimal in any case).

The hit part is what I am unfamiliar with. I dont know for sure but having a skull boss mob compared to a regular 73 mob, you end up with different hit tables (at least it seems that way). As such I didnt think that the extra miss chance on boss mobs came from extra defense but rather something attributed to the fact that it was a skull mob in and of itself. From what you have written it would appear that the extra miss chance (that brings it to 8.6 %) is due to odd mechanics of defense vs. weaponskill for higher level mobs. If ya have any reference for that I'd certainly appreciate reading it since it makes weapon skill far more valuable than I had considered.

Nor am I modeling a two-roll system for either yellow or white attacks; maybe we should see what the rogues have decided with regards to that at some point in the near future.

Ok, an increase in hit will increase your damage output by a fixed percentage; however, because percentages are scalars, this will increase the utility of your other stats (just how having no hit ~decreases~ the utility of your stats by 8.6% simply because 8.6% of the time you miss and a miss with a billion AP is the same as a miss with 0 AP).


Point.

Omen was just an example; you could be attempting to proc ~anything~ really. My point was more that missing makes nothing proc whereas a hit means you have a chance to proc something.


I don't and my valuation of hit is still higher than yours and I really don't think that my math is simply that horrible; I may have a formula wrong somewhere but I'm fairly sure that I'm not all that far off of the mark.

Statistics. If you want to know how likely it is that a random event will occur over a sequence of effects that could trigger it, you go about it in the following manner:

• Take the %Chance and generate the likelihood of the event NOT occurring (1-%Chance).
• Chance to not Occur over N Hits = (1-%Chance)^N
• Chance to Occur over N Hits = (1-Chance to Not Occur)

Thus, over 5 hits, we had a 36% chance to miss once, reducing the chance to proc OOC once, which was ~8%, by that amount (yielding an expected procrate of 5.1%).

It all depends on how exactly +Skill is modeled; again, we should probably refer to rogue spreadsheets to find out how they've decided to do it so that we can save time on that debate.

That's something else I haven't taken into account; still, it costs a good deal less and does it really do 50% less? Are you taking the lost hit into account when you turn a hit into a crit?

Hit (100%) -> Crit (210%(?)) yields a 110% increase in damage.
Miss (0%) -> Hit (100%) yields a 100% increase in damage.

So if hit costs 29% less and does 90.91% of the damage, that's pretty good in my book.

Nor can I to be honest; just presenting my PoV on the situation.

Pred Instincts states that it increases the bonus damage by 10% (as opposed to the net damage).


22 from my pair of Shattrath leggings (legguards?) (either 5.58 Skill or 5.8 skill; can't remember which).

Pretty much; I'm glad our models agree on ~something~ ;D.

As a side note, I wonder if our differing valuations for the stats are generated from using a different testing methodology & definitions (and my generally weaker tests, which are pretty much doomed to have poorer accuracy by definition); for example, I refuse to call something a 'cycle' unless I start and end at roughly the same point, especially when energy & debuffs are concerned whereas everyone else seems to use the 1 Mangle, 3 Shreds, & Rip 'cycle' (with excess energy just kinda going into the void?).

This is not correct. The text from Predatory Instinct reads: While in Cat Form, Bear Form, or Dire Bear Form, increases your damage from melee critical strikes by 2% [per point] and your chance to avoid area effect attacks by 3% [per point]. Originally it had SAID it only increased the bonus damage but it always functionned by increasing the total damage. One or two patches ago they updated the tooltip to the text I posted above.

Usually for those doing a mangle/3 shred/rip cycle end up dividing by the total energy used for this to determine DPS. As such the "extra" energy isnt wasted or anything its just taken into account in terms of decimals. Due to the fact that varying crit rates and different energy returns on a miss/dodge, finding an optimal true cycle seems extremely difficult and far too time consuming since it would need to be redone for every different crit/hit chance.

Due to the complex nature of modelling the yellow attacks I think a simulator using a RNG and thousands of seconds of attacks is probably a simpler way to get the desired results. I dont have any experience coding such things except in basic C so I wont be able to make a pretty interface or anything for it but I am working on a program to do this so hopefully I can get that done soon and post the code which may help some.

Ah; wowhead is still using the "Bonus Damage" tooltip. I did model a *2.2 crit though.

Anyway, a hit still nets 83.3% of a crit, damage-wise. 16.7% less damage for 29% less cost works for me .



Yeah, coding GUIs isn't my thing either (I can design some good ones but hate coding & tweaking them). I may code up something in the next few days(/weeks) but we'll just have to see.

For white damage, no doubt, hit is better, point for point, than crit and that shows it. Its just for yellow where it the opposite is true due to not losing the full amount of energy on a miss (except on finishers).

? Which one are you referring to as the 'generally accepted best'? Surely not the Haste one? Relentless blows every meta I've used out of the water, hands down.

As for the topic, I found this kind of odd:

http://www.juggernautguild.com/wws/2...13/arkadu.html

My tooltip says 8.86 chance to hit. I didn't get a single miss the whole fight, until the instant she transformed into the Void Walker, and immediately got 5 consecutive misses; 3 melee, 2 shreds. I have no explanation for this, aside from my theory that NPCs have dynamic combat ratings, but even then, I don't see how that could have affected my miss rate like that.

You're right I overlooked the extra crit damage from talents/meta.
And obviously if you come up with 1% hit being worth 50% less than 1% crit then it's definitely not worth getting over crit with equally geared items ;p but the people posting earlier didn't even think about figuring out such values.

As for procs from +hit, this is why I neglect them:
Let's say you have a proc that does a whopping 5% of your DPS. Increasing +hit by 1% will give you roughly 1% more procs (a little more since you'll never really be at 100% hit and if you are discussion of effects of +hit is moot, but it's roughly 1% because for example 91/90=>1.11% increase). Increasing 5% of your DPS by 1% is 0.05% DPS. Not only 0.05% DPS is nothing, you probably already might've took the proc's DPS into account anyway when modeling your white damage... And the reason I can say 0.05% DPS is nothing is that comparing to my mage 1 whopping spell damage gives me 0.05% DPS increase... So with the perfect model you want to take this into account but when showing people why they're looking at the hit VS crit wrong, you can safely neglect procs.

All in all, there seem to be enough spreadsheets around to help you see which stats are good for you, so no need to bring silly arguments for why 1 stat is better or worse than the other.

Well as I posted earlier suppose you have a situation where *some* amount of hit will allow 2 hits and if instead we have crit (same %) we get 1 crit and 1 miss.

In this case, with mangle we get:

2 Hits: 2x mangle damage, 2 combo points, 2 chances at procs, 80 energy used
1 crit 1 miss: 2.2x mangle damage (2.26 with relentless earthstorm), 2 combo points, 1 chance at procs, 48 energy used.

Neglecting to the procs for a moment we have:

2*dmg / 80 energy vs. 2.2*dmg/48 energy

0.025*dmg/energy vs. 0.0458 dmg/energy

which means hit is worth 54.58% as much as crit here since as was shown in a previous post, damage is energy limited not time limited (so even if the miss makes you wait another energy tick, that extra 20 energy is not LOST but will be used at the next time). Even if we add procs here, it seems unlikely that it will make up the difference despite hit costing much less than crit.

Yup, and I model omen. If you have other things that proc off hit, then these need to be modeled. Some classes have items that proc off misses. The modeling for these belong on the item itself, and +hit will show an increased benefit if you have them in your gear set.

Right, what i quoted was a 10% raw reduction in damage plus a 36% reduction in OOC procs.
The raw reduction in white damage is under 7%, and the raw reduction in yellow damage is around 2%, so 10% is off by more than a factor of 2.
A 40 energy attack can be OOC'd by the 4 preceding white hits or the 1 preceding yellow, overlaps give no benefit. With a 3% proc chance the chance of it being OOC'd is 1-0.97^5 = 11.4%, with 8.4% to miss the proc chance becomes 2.748% (OOC and hit are independent i.e. 2 roll) giving an 11.2% chance of being OOC'd. The 0.2% difference is a 7.9% difference in OOC procs, that is, 1% hit scales extra damage to OOC by a little under 1%. Apologies if you didn't mean 36%, but it's what you said

Yes.

It's 120% not 110%, and that's white damage only. With yellow damage, crits give you combo points, and hits give you very little (100% for 40 energy vs 0% for 7 energy. Think of it as a bet, but if you win you bet 40 energy and if you lose you only bet 7).

There isn't any excess energy, because your cycle is over as soon as your energy has let you perform all the moves. Having rips overlap by enough that you lose energy ticks if you wait is rare, and an extra shred is 'close enough' to your average DPE in such cases. (and thus doesn't affect your DPS because energy is used to measure time).
If you're talking about the 5 energy leftovers, we're modelling a discrete probabilistic system continuously, so we assume continuous energy regen except when there's good reason not to. Since the exact timing of attacks mostly doesn't matter, and you've got flexibility to delay attacks to maintain the relationships you do need, this works as long as we don't use a cycle that assumes 11 shreds in 12 seconds or equal silliness.

I'd suggest a spreadsheet that models white DPS as damage-per-attack and yellow dps as damage-per-energy. Oh wait... =P

Haha, good catch - I was referring to the haste one which has been niggling at the back of my mind - I got relentless and then thought I read thundering was better. I think I got my druid/hunter threads confused.
That astromancer scenario is definitely weird. I can't think of an explanation, and I've never seen the encounter so I won't speculate. (EDIT: other than to say I don't know of any mechanic other than a visible debuff on the player, or a defense increase on the mob, that increases the player's chance to hit the mob. To give a 1% chance of missing 5 attacks in a row when you're hit capped would require a ~1000 defense increase. There's been suggestions in the past that the combat log returns dummy results when something weird happens server side. Now i'm speculating.)

Definitely, and I've run into a few more thanks to this discussion. I guess I didn't understand the logic behind the assumption that you should be hit capped. The answer seems to be threefold
* it's partly an unfounded assumption (I think this is important for people to know)
* it helps you manage your cycles more accurately, which often helps dps more than a spreadsheet would show
* +hit isn't as terrible as I thought it was, although it is the worst of the common DPS stats.

I wish they would change the system to be that logical

Calculate the chance of the attack right before you run out will miss and then calculate how much a certain amount of hit rating will help that chance... Then compare with other stats that work all the time.
Yes if you miss the first attack right before an energy tick you'll lose an energy tick instead of whatever you were supposed to lose. If you miss right before the mob dies and you can't attack another before your energy fills back up you lost the attack. And I bet you can come up with some more rare examples of where missing is worse than what it was made to be above. But those cases are few and insignificant in the long run if they even exist in the fight you're working on beating.

So do we have somewhat of a consensus on how to weight each value in terms of dps increase? It looks like it's shaping out to be something like this at around 3000 AP.

AGI = 2.5
STR = 2.4
AP = 1
CR = 0.8
Haste = 0.7
Hit = 0.7

haste is all find and good, but I'll let the rogues have it while I ninja all the loot I really want. :]

next thought, FCS reduces the dodge chance of bosses as well. I cannot currently find the link to the post about the differing % reductions to miss, dodge and block on increased weapon skill however :/

Even if +crit rating is very marginally better than +hit, I hate, loathe, despise and generally dislike seeing he words 'miss' 'dodge' 'parry' and 'block' appear when i'm trying to kill something.

Final thought: Hit has a cap. crit and agi don't. Setting and achieving goals makes for happy campers.

That doesn't look to be factoring in Blessing of Kings and the 3% to all stats talent.

Assuming the above.
STR would then be 2.72 (2.4 * 1.1 * 1.03)
AGI 2.83 (2.5 * 1.1 * 1.03)

Given two items of nearly the same dps from the above, I'll favor one with hit versus one without hit (until close to the cap).

Its not always about losing /gaining dps, its about personal preference choices when the dps difference is minor.

A link of all items sorted with the above values (top 12):
http://www.lootzor.com/index.php?c=1...u.8hedu.7heeu1

Of course it depends if you have kings or not, and in most situations where it matters you will. Bufferd only with kings and my own MOTW i have around 3000AP.

AGI = 2.88
STR = 2.70
AP = 1.00
CR = 1.77
Haste = ??? (I still don't model this yet)
Hit = 1.26
Weapon = 12.33 (for things like the Crystalforged Trinket and ring enchants)

Very well put. The predictability effect is hard to quantify but it's definitely there.

I don't want to factor in kings. Why isn't weapon damage rated at 14 out of curiosity, because it doesn't add rip damage? Does the natural weapons bonus apply before or after +damage bonuses? Sadly I don't have a crystalforged myself to look at a paperdoll.